At some point you have heard the phrase. You have probably said it yourself. Nobody is irreplaceable. It comes out after a resignation, when someone needs to project calm, or when a manager wants to quietly signal to a high performer that the leverage doesn’t actually favor them. It sounds reasonable.

I want to pull on it a little, because I think it does real damage when organizations actually build around it.

In 2012, two researchers, Ernest O’Boyle and Herman Aguinis, looked at performance data across 633,263 people: researchers, entertainers, politicians, athletes. They were trying to settle what sounds like an administrative question until you see the implications. Does human performance follow a bell curve, where most people cluster near average and outliers are rare on both sides?

It does not.

Instead of a hill with most people near the middle and a few at each end, the shape is a steep cliff. Almost everyone is clustered at the bottom, and a very small number are so far above the rest that the gap barely fits on the same chart. A McKinsey analysis drawing on that same work put a number to it: the top five percent of workers outperform average workers by 400 percent. Not 40. Four hundred. The finding held across industries, job types, and time frames.

The cost of losing one of those people is not just a hiring inconvenience. Research from the Center for American Progress found that replacing a highly trained employee can cost up to 213 percent of their annual salary, and it can take up to two years for a replacement to reach the same level of output as the person who left. Research by SAP and Oxford Economics found that less than half of high performers are satisfied with their jobs, and one in five say they are likely to leave within six months. These are the people producing 400 percent more than the person next to them on the org chart, and most organizations are managing their retention with the same philosophy they apply to everyone else.

The bell curve assumption sitting inside salary bands, performance reviews, and headcount planning is not a description of reality. It is a description of what organizations find more comfortable to believe.

Takeaway: “nobody is irreplaceable” reflects how organizations prefer to see their people, not how performance actually distributes.

Apple lost Steve Jobs twice. The first time, in 1985, the company continued. It had products, teams, and real momentum from what he had already built. Then over the following decade it failed to produce anything genuinely new, lost its desktop publishing margins when IBM undercut on price, and by the mid-1990s was losing over a billion dollars a year.

Jobs returned as interim CEO in 1997. Within twelve months Apple moved from a $1.04 billion loss to a $309 million profit.

The easy interpretation is that he was a strong executive who corrected what had drifted. The more precise description is that he carried something back that the organization had no way of preserving without him: a specific combination of taste, obsession, and willingness to commit to things that had no proven market. When he encountered a designer sitting with a pile of prototypes, one of which would become the iMac, the decision to go all-in on that vision did not come out of a product committee. It came from one person who could see what the thing was before the data could tell anyone.

After his death in 2011, Apple retained the brand, the revenue base, and the product lines he had built. What it did not retain was the capacity to conceive entirely new categories or to take the kind of risks that produce them. The momentum lasted years. The category-defining product innovation largely stopped.

Takeaway: Irreplaceability is not about seniority or legend. It is about a person whose specific combination of capabilities produces something the organization genuinely cannot produce without them.

There is a seductive argument in most business books that runs roughly like this: great companies are built on great systems, not great people. The truly mature organization is one where any seat can be filled without disruption, where process and institutional memory carry the knowledge forward.

This is not an argument against process. Algorithmic output is genuinely extraordinary. It is efficient, reduces errors, runs at scale, and does not have bad days. Watching a well-designed system operate is one of the more satisfying things you can witness inside a company.

The problem is when organizations fall so in love with their own systems that they stop caring about the people feeding them. Think of the most sophisticated kitchen ever built: calibrated temperatures, precise timers, spotless technique. It can still produce food nobody wants to eat. The kitchen does not have taste. It does not know what the season is, what people are hungry for, or what combination of things nobody has tried yet. It executes, beautifully, whatever you put into it.

The irreplaceable people are the ones deciding what goes in. More critically, they are the ones who can read where things are heading before the data confirms it, adjust in response, and do it again when conditions shift. Process without that just repeats what already worked, which is fine until the market moves and the system keeps producing the same thing.

Over time, organizations that invest so heavily in the kitchen that they stop investing in the cooks end up with a very efficient operation that gradually stops meaning anything.

Takeaway: The process is only as good as the people deciding what it is for.

Companies built almost entirely on replaceable people do exist and some are enormous. But something else is carrying the weight: a patent portfolio competitors cannot legally cross, a technology moat that is genuinely difficult to replicate, or scale that makes competition economically irrational. These are real advantages. They are also time-buying advantages, not growth-generating ones.

IBM had all three at its peak. So did Nokia. So did Kodak, which in 1975 had one of its own engineers, Steve Sasson, build the first working digital camera. The organization looked at what Sasson had made, concluded it threatened the film business the entire company existed to protect, and shelved it. The process asked whether the invention fit the existing strategy. It did not. Kodak filed for bankruptcy in 2012.

What Kodak ran out of was not process. It was people with the capacity and authority to recognize what they were looking at and act on it regardless of what the existing model required.

Takeaway: a moat buys time. When it narrows, and it always does, what you have left is whoever you built the organization around.

Frequent readers will be familiar with a formula I use to navigate relationships that resist plain language. Value, I have argued, is proportional to existing demand multiplied by relevance, divided by friction. The equation stretches across different scenarios, and I want to augment it here to describe what actually happens when organizations choose between building around exceptional people or replicable process.

Output ∝ (Irreplaceable people × Size of challenge) / Friction

You read this as: an exceptional person handed a small problem inside a slow-moving organization produces roughly the same result as a capable but unremarkable one. The person is not always the limiting factor. Often it is the scale of what they have been asked to do, or the weight of process sitting on top of them. Give that same person a genuinely consequential problem with enough room to pursue it and the returns stop being linear.

This is also why compensation is not a purely operational question. Someone who is genuinely irreplaceable is constantly running an implicit calculation on the gap between what they produce and what they receive. That calculation resolves eventually, and not usually in favor of patience.

If your people can be swapped out without meaningful loss, your company can be replaced too. Not today, not without real effort from whoever is trying, but the conditions for it are already in place. What you have built is a replicable system, and replicable systems attract competition that undercuts on cost, automates pieces away, or simply waits for the structural advantages to expire.

The “nobody is irreplaceable” line is very convenient for the people it protects. A company that genuinely accepts the reality has to act on it, pay accordingly, and have honest conversations about what certain people actually produce versus what they receive. Believing the myth makes all of that disappear. It keeps pay scales tidy, keeps managers comfortable, and keeps the exceptional person from knowing exactly how much of the operation runs on them.

So find the irreplaceable ones. Pay them well. Give them something genuinely worth building. This is the easy part to say and the hard part to do, because exceptional people are not always easy to identify. They come in different forms, and recognizing them requires a specific kind of judgment that you either have or you are still developing.

If you cannot see exceptional people for what they are, you will not know when one is sitting across from you. You will evaluate them against the wrong criteria, lose them to someone who understood what they were looking at, and console yourself afterward with the idea that nobody is irreplaceable anyway.

The ability to recognize irreplaceable people is itself irreplaceable.

STRATEX by Naz explores the cognitive and behavioral foundations of how people, organizations, and markets work.